Bacterial Analogs of Plant Tetrahydropyridine AlkaloidsMediate Microbial Interactions in a Rhizosphere Model System

Gabriel L. Lozano,a,b Hyun Bong Park,c,d Juan I. Bravo,a,b Eric A. Armstrong,e John M. Denu,e Eric V. Stabb,f

Nichole A. Broderick,g Jason M. Crawford,c,d,h Jo Handelsmana,b

aWisconsin Institute for Discovery and Department of Plant Pathology, University of Wisconsin–Madison, Madison, Wisconsin, USAbDepartment of Molecular, Cellular and Developmental Biology, Yale University, New Haven, Connecticut, USAcDepartment of Chemistry, Yale University, New Haven, Connecticut, USAdChemical Biology Institute, Yale University, West Haven, Connecticut, USAeWisconsin Institute for Discovery and Department of Biomolecular Chemistry, University of Wisconsin–Madison, Madison, Wisconsin, USAfDepartment of Microbiology, University of Georgia, Athens, Georgia, USAgDepartment of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut, USAhDepartment of Microbial Pathogenesis, Yale School of Medicine, New Haven, Connecticut, USA

ABSTRACT Plants expend significant resources to select and maintain rhizospherecommunities that benefit their growth and protect them from pathogens. A betterunderstanding of assembly and function of rhizosphere microbial communities willprovide new avenues for improving crop production. Secretion of antibiotics is onemeans by which bacteria interact with neighboring microbes and sometimes changecommunity composition. In our analysis of a taxonomically diverse consortium fromthe soybean rhizosphere, we found that Pseudomonas koreensis selectively inhibitsgrowth of Flavobacterium johnsoniae and other members of the Bacteroidetes grownin soybean root exudate. A genetic screen in P. koreensis identified a previously un-characterized biosynthetic gene cluster responsible for the inhibitory activity. Metab-olites were isolated based on biological activity and were characterized using tan-dem mass spectrometry, multidimensional nuclear magnetic resonance, and Mosherester analysis, leading to the discovery of a new family of bacterial tetrahydropyri-dine alkaloids, koreenceine A to D (metabolites 1 to 4). Three of these metabolitesare analogs of the plant alkaloid �-coniceine. Comparative analysis of the koreen-ceine cluster with the �-coniceine pathway revealed distinct polyketide synthaseroutes to the defining tetrahydropyridine scaffold, suggesting convergent evolution.Koreenceine-type pathways are widely distributed among Pseudomonas species, andkoreenceine C was detected in another Pseudomonas species from a distantly relatedcluster. This work suggests that Pseudomonas and plants convergently evolved theability to produce similar alkaloid metabolites that can mediate interbacterial com-petition in the rhizosphere.

IMPORTANCE The microbiomes of plants are critical to host physiology and devel-opment. Microbes are attracted to the rhizosphere due to massive secretion of plantphotosynthates from roots. Microorganisms that successfully join the rhizospherecommunity from bulk soil have access to more abundant and diverse molecules,producing a highly competitive and selective environment. In the rhizosphere, as inother microbiomes, little is known about the genetic basis for individual species’ be-haviors within the community. In this study, we characterized competition betweenPseudomonas koreensis and Flavobacterium johnsoniae, two common rhizosphere in-habitants. We identified a widespread gene cluster in several Pseudomonas spp. thatis necessary for the production of a novel family of tetrahydropyridine alkaloids thatare structural analogs of plant alkaloids. We expand the known repertoire of antibi-

Citation Lozano GL, Park HB, Bravo JI,Armstrong EA, Denu JM, Stabb EV, BroderickNA, Crawford JM, Handelsman J. 2019.Bacterial analogs of plant tetrahydropyridinealkaloids mediate microbial interactions in arhizosphere model system. Appl EnvironMicrobiol 85:e03058-18. https://doi.org/10.1128/AEM.03058-18.

Editor Marie A. Elliot, McMaster University

Copyright © 2019 Lozano et al. This is anopen-access article distributed under the termsof the Creative Commons Attribution 4.0International license.

Address correspondence to Jason M. Crawford,jason.crawford@yale.edu, or Jo Handelsman,jo.handelsman@wisc.edu.

Received 20 December 2018Accepted 2 March 2019

Accepted manuscript posted online 15March 2019Published

GENETICS AND MOLECULAR BIOLOGY

crossm

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 1Applied and Environmental Microbiology

2 May 2019

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

otics produced by Pseudomonas in the rhizosphere and demonstrate the role of themetabolites in interactions with other rhizosphere bacteria.

KEYWORDS Flavobacterium johnsoniae, Pseudomonas koreensis, antibiotics, bacterialcompetition, convergent evolution

Plants were long thought to be defined by their genes and environments. It hasrecently become apparent that plants are also shaped by their microbiomes, i.e.,

the communities of microorganisms that live on, around, and inside them (1). Micro-biomes modify many environments, including humans, animals, oceans, soils, and hotsprings. Comprehensive investigations of the interactions between microbiomes andtheir environments, as well as the interactions within microbiomes that contribute totheir function and stability, are important to understanding diverse niches on Earth,including those associated with plants.

The rhizosphere comprises plant root surfaces and their surrounding soil microen-vironments. Bacteria are attracted to these environments by the massive amount ofplant photosynthate, in the form of sugars, organic acids, and amino acids, which issecreted from roots (2). Bacteria that colonize the rhizosphere play an essential role inplant growth and resistance to pathogens. For example, some members secrete planthormone-like molecules, such as indole acetic acid, gibberellic acid, cytokinin, andabscisic acid, that promote plant growth (3), whereas others suppress plant diseases bysecreting diverse compounds such as zwittermicin A, 2,4-diacetylphloroglucinol, andpyoluteorin (4). Thus, bacterial rhizosphere communities represent a rich reservoir ofbioactive metabolites.

Use of bacteria for biological control of plant disease has been pursued for decades,but foreign microorganisms typically do not persist in native rhizosphere communities(5). Nutrient abundance, host availability, and microbial interactions define indigenousmicrobial community structures and limit colonization by invading bacteria. To engi-neer plant microbiomes to improve agricultural systems, a better understanding of theinterbacterial interactions that dominate the rhizosphere is needed.

We developed the hitchhikers of the rhizosphere (THOR), a model system toexamine the molecular interactions among core bacterial members of the rhizosphere(6). This model system is composed of Bacillus cereus, Flavobacterium johnsoniae, andPseudomonas koreensis, which belong to three dominant phyla within the rhizosphere,Firmicutes, Bacteroidetes, and Proteobacteria, respectively. The three members displayboth competitive and cooperative interactions. For example, P. koreensis inhibitsgrowth of F. johnsoniae but not in the presence of Bacillus cereus. Inhibition was onlyobserved when bacteria were grown in soybean root exudate and is specific formembers of the Bacteroidetes, based on testing a collection of taxonomically diverserhizosphere bacteria (6). In this study, we characterized the genetic and molecularmechanisms by which P. koreensis inhibits F. johnsoniae. We determined that a newfamily of bacterial tetrahydropyridine alkaloids, designated koreenceine A to D (com-pounds 1 to 4), are produced by an orphan polyketide synthase (PKS) pathway andmediate inhibition of members of the Bacteroidetes. Koreenceine A, B, and C arestructural analogs of the tetrahydropyridine alkaloid �-coniceine and its reducedpiperidine alkaloid coniine, produced by plants, and comparisons of the plant andbacterial biosynthetic pathways support a convergent evolutionary model.

RESULTSIdentification of an orphan P. koreensis pathway that is responsible for inhib-

iting growth of F. johnsoniae. To identify the genes required for inhibition of F.johnsoniae by P. koreensis in root exudate, we screened 2,500 P. koreensis transposonmutants and identified sixteen that did not inhibit F. johnsoniae (Table 1). Two of thesemutants mapped in BOW65_RS02935 and BOW65_RS02945, which are part of anuncharacterized polyketide biosynthetic gene cluster containing 11 genes (Fig. 1). Wedeleted the entire gene cluster (kecA-kecK::Tn), which abolished inhibitory activity

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 2

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

against F. johnsoniae and other members of the Bacteroidetes (see Fig. S1 in thesupplemental material). We designated this pathway an orphan pathway, since theencoded natural product was unknown.

We developed a defined medium in which P. koreensis produces the gene cluster-dependent inhibitory activity against F. johnsoniae that was observed in root exudate(Fig. S2A). Since we identified two independent mutants in the gene encoding a sensorhistidine kinase, cbrA, that was required for activity in root exudate (Table 1), wedeveloped a defined medium with the goal of activating the CbrAB system (7), whichcontrols the utilization of alternative carbon sources such as amino acids (8). Adding todefined medium the same mix of amino acids that was used to supplement rootexudate induced P. koreensis to produce inhibitory activity in the defined medium (Fig.S2A). We next tested 19 amino acids and identified five, including aspartate, thatinduced inhibition of F. johnsoniae by P. koreensis (Fig. S2B). A nonhydrolyzable analogof aspartate, N-methyl-DL-aspartate (Asp*), did not stimulate inhibitory activity, suggest-ing that catabolism of certain amino acids is required for activity (Fig. S2B).

Characterization of koreenceine metabolites from the orphan P. koreensispathway. To characterize the inhibitory metabolites from the orphan P. koreensispathway, we compared the metabolomes of the wild-type strain and the noninhibitorymutant grown in root exudate. High-performance liquid chromatography-mass spec-trometry (HPLC-MS)-based analysis of the crude organic extracts led to the identifica-

TABLE 1 P. koreensis mutants identified in the genetic screen with loss of inhibitory activity against F. johnsoniae

Mutant ID Predicted function Class

1 BOW65_RS02945 Pyridoxalphosphate-dependent aminotransferase Secondary metabolite production2 BOW65_RS02935 3-Oxoacyl-ACP synthase Secondary metabolite production3, 4 BOW65_RS24575 Two-component sensor histidine kinase CbrA Cell signaling and transcription regulation5, 6 BOW65_RS20110 CysB family transcriptional regulator Cell signaling and transcription regulation7 BOW65_RS21410 Multisensor hybrid histidine kinase Cell signaling and transcription regulation8 BOW65_RS06475 Outer membrane protein assembly factor BamC Cell surface9 BOW65_RS22455 Peptidoglycan-associated lipoprotein Cell surface10 BOW65_RS29255 Phospholipid/glycerol acyltransferase Cell surface11 BOW65_RS28295 Acetylglutamate kinase Metabolism12 BOW65_RS08620 Methylcitrate synthase Metabolism13 BOW65_RS24475 Ketol-acid reductoisomerase Metabolism14 BOW65_RS07790 Succinyl-CoA synthetase subunit alpha Metabolism15 BOW65_RS08625 2-Methylisocitrate dehydratase Metabolism16a BOW65_RS24600 3-Methyl-2-oxobutanoate hydroxymethyltransferase Metabolism16a BOW65_RS24605 Pantoate-beta-alanine ligase MetabolismaTransposon insertion in the promoter 5= region of an operon conformed for these two genes.

FIG 1 Koreenceine biosynthetic locus and the predicted function of each gene. Black arrows indicatelocations of the transposons of the mutants identified.

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 3

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

tion of peaks 1 to 4 that were completely abolished in the mutant (Fig. 2). We carriedout bioassay-guided preparative-scale HPLC fractionation of the crude organic extractfrom a culture (5 liters) of the wild-type P. koreensis grown in defined medium. Peaks 1,2, and 4 were detected in fractions with antimicrobial activity against F. johnsoniae.High-resolution electrospray ionization-quadrupole time of flight MS (HR-ESI-QTOF-MS)data for peaks 1 to 4 revealed m/z 208.2067, 210.2224, 226.2171, and 278.1885, allowingus to calculate their molecular formulas as C14H26N, C14H28N, C14H27NO, andC14H29ClNO2, respectively (Fig. 2 and Fig. S3). We then proceeded with mass-directedisolation of these compounds from a larger-scale culture in defined medium (12 liters)of wild-type P. koreensis for nuclear magnetic resonance (NMR)-based structural char-acterization.

The chemical structures of compounds 1 to 4 were characterized through 1H,two-dimensional (2D)-NMR (gradient correlation spectroscopy [gCOSY], gradient het-eronuclear single quantum coherence [gHSQC], and gradient heteronuclear multiple-bond coherence [gHMBC]), tandem MS, and Mosher ester analysis (Fig. 3 and Fig. S3 toS7). Briefly, 1H NMR spectra combined with gHSQC of peak 2 revealed the presence ofsix well-defined methylene groups, including one downfield-shifted signal, six addi-tional overlapped methylene groups, and one methyl group. Consecutive COSY cross-peaks from a triplet methyl H-15 (chemical shift [�H], 1.86) to a methylene H-7 (�H, 2.53)established a partial structure of a nonane-like hydrocarbon chain. Additional COSYcorrelations from a downfield-shifted methylene H-2 (�H, 3.53) to a methylene H-5 (�H,2.72) also constructed a shorter 4� CH2 chain. Key HMBC correlations from H-2, H-4,and H-7 to C-6 allowed us to construct the tetrahydropyridine core in compound 2. Incontrast, the 1H NMR spectrum of compound 4 showed the presence of a hydroxylmethine H-3 (�H, 3.94), which was evident by COSY correlations with both methyleneH-2 and H-4. The connectivity between H-1= (�H, 3.20) and H-4= (�H, 3.59) was estab-lished by additional COSY correlations, which was further supported to be a4-chlorobutanamine-like partial structure by the presence of a monochlorine isotopedistribution pattern in the HR-ESI-QTOF-MS data. HMBC correlations from H-2 and H-1=to an amide carbon, C-1, unambiguously constructed the chemical structure of com-

FIG 2 Extracted ion chromatograms from LC-HR-ESI-QTOF-MS of koreenceine A to D for the wild type andkecA-kecK deletion mutant.

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 4

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

pound 4 to be N-(4-chlorobutyl)-3-hydroxydecanamide. Modified Mosher’s reaction onthe secondary alcohol at C-3 determined the absolute configuration of C-3 to be R,completing the absolute structure of compound 4. The structure of compound 1, ananalog of compound 2, was elucidated based on the 1H and COSY NMR data thatindicate the position of a trans-double bond between H-7 and H-8. Finally, the chemicalstructure of compound 3 was deduced by comparative high-resolution tandem MSanalyses with the closely related compounds 1 and 2.

Koreenceine structure-activity analysis. We estimated the MICs of both koreen-ceine B (compound 2) and D (compound 4) to be 200 �g ml�1 against F. johnsoniae. Wepredicted that koreenceine D does not have a major role in the inhibitory activity, sincekoreenceine D (compound 4) is present in root exudate cultures at levels 100 times lessthan that of koreenceines A (compound 1), B (compound 2), and C (compound 3) (Fig.2). We could not estimate a MIC for koreenceine A, as its levels diminish during thepurification process. We synthesized koreenceine A and observed similar decomposi-tion during purification (data not shown). Thus, we tested a semipurified fraction ofkoreenceine B with a trace of koreenceine A, which had a stronger inhibitory effect than

FIG 3 Structural characterization of koreenceines. (A) Chemical structures of compounds 1 to 4. (B)Key COSY and HMBC NMR correlations of compounds. (C) Δ�S-R (in ppm) for the MTPA esters ofcompound 4.

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 5

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

koreenceine B alone (MIC of 40 �g ml�1). The significant increase in activity associatedwith trace amounts of koreenceine A suggests that this molecule is the major inhibitorymolecule against F. johnsoniae in the THOR rhizosphere model or is synergistic withkoreenceine B.

Proposed biosynthesis of koreenceine metabolites. �he defining tetrahydropyri-dine core of koreenceine metabolites A to C is observed in plant alkaloids such as�-coniceine, a well-characterized alkaloid from poison hemlock (Conium maculatum)(9). Examination of the genes in the koreenceine biosynthetic gene cluster identifiedencoded proteins with homology to predicted or previously identified enzymaticactivities needed for the production of �-coniceine in plants (10–13) (Fig. 1 and 4). Wepropose the following biosynthetic pathway of koreenceine A to C. The first five genesof the cluster, kecABCDE, encode a type II polyketide synthase system: kecA encodes anacyl carrier protein (ACP), kecB and kecD encode �-ketoacyl synthases (KS�), and kecCand kecE encode partial �-ketoacyl synthases with conserved thiolase domains (chainlength factor [CLF] or KS�). This cluster may encode production machinery for two-heterodimer systems, KecB-KecC and KecD-KecE, for polyketide elongation with KecAand likely participate in the formation of a triketide intermediate derived from thecondensation of two malonyl units and a decanoyl, 3-hydroxy-decanoyl, or trans-2-decenoyl unit (Fig. 4, step 1). �-Keto reductive modifications could be catalyzed byKecG and KecH reductases (Fig. 4, step 2). Aminotransferase KecF is predicted tocatalyze transamination of the aldehyde intermediate facilitating tetrahydropyridinecyclization. KecF appears to be a multidomain protein with a predicted aminotransfer-ase at the N terminus and a general NAD(P)-binding domain (EMBL-EBI accession no.IPR036291) and a conserved protein domain, COG5322, at the C terminus. Interestingly,long-chain fatty acyl-ACP reductases from Cyanobacteria share these features and

FIG 4 Predicted biosynthetic pathway for �-coniceine in plants and proposed biosynthetic pathway forkoreenceine B in P. koreensis. Similar functions are color coded to highlight the similarity between theplant pathway and the koreenceine biosynthetic locus.

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 6

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

generate fatty aldehydes from the reduction of fatty acid intermediates bound to ACP(14). We predict that KecF reduces the ACP-polyketide intermediate to a polyketidealdehyde with the C-terminal domain (Fig. 4, step 3), followed by transamination by theN-terminal domain (Fig. 4, step 4). Finally, the amine intermediate could undergo anonenzymatic cyclization, as observed in �-coniceine (Fig. 4, step 5) (12). We predictthat koreenceine D is derived from koreenceine C by an unidentified halogenasereaction, as analogous metabolite sets have been detected in plants (Fig. S8) (15). Thelast three genes, kecIJK, may participate in the translocation of the koreenceine alka-loids outside the cell. KecI and KecJ are hypothetical proteins predicted to localize inthe membrane, and KecK has homology with membrane-bound drug transporters.

Convergent evolution of pathways for production of �-coniceine-like alkaloidsin plants and P. koreensis. The biosynthetic pathway for production of �-coniceine isstill under investigation, but 14C feeding experiments in C. maculatum coupled withchemical degradation of the labeled products suggest that �-coniceine is not derivedfrom an amino acid like other plant alkaloids; rather, it is derived from a polyketidechain produced by the condensation of acetate units (10). Type III polyketide synthasescommon in plants are iterative homodimers that orchestrate the acyl-CoA-mediatedpriming, extension, and cyclization reactions for polyketide products without the use ofacyl carrier proteins (16). Recently, Hotti et al. found CPKS5, a non-chalcone synthase/stilbene synthase (CHS/STS)-type III polyketide synthase expressed in tissues thatcontain �-coniceine (Fig. 4) (11). The pathway that we identified in P. koreensis encodestwo type II PKS systems involved in production of the proposed polyketide interme-diate (Fig. 1). Although the pathway that we identified in P. koreensis producescompounds related to the plant alkaloids, the PKSs from the plant and bacterialkingdoms share little similarity. We propose that convergent evolution led to twodifferent polyketide pathways for the production of �-coniceine-like metabolites inplants and bacteria.

Distribution of the koreenceine cluster. Similar koreenceine-like clusters havepreviously been identified by functional screens for antimicrobial activities (17, 18);however, there are no reports of the metabolites being produced. We identified 179koreenceine-like clusters in genomes in the NCBI database (June 2018). The majority ofthese clusters are in Pseudomonas genomes, although we found some partial clustersin Xenorhabdus and Streptomyces species genomes (Fig. 5A). We used maximum-likelihood analysis of the amino acid sequence of the aminotransferase-reductaseprotein, KecF, as a representative of the koreenceine cluster for phylogenetic recon-struction. We observed four main clades that are each associated with a bacterial genus(Fig. 5A). Clades A and B, which contain 93% of the clusters, are found in Pseudomonasgenomes. The koreenceine gene cluster identified in P. koreensis in this study belongsto clade A; other clade A clusters are located in the same genomic context in P.koreensis and P. mandelii, two closely related species in the P. fluorescens complex (19)(Fig. 5B). Clades C and D were found in Streptomyces and Xenorhabdus spp., respec-tively, although the kecI, kecJ, and kecK genes are missing.

Clade A and the P. koreensis and P. mandelii genomes are phylogenetically parallel(Fig. 5C). We hypothesize that the gene cluster was acquired before Pseudomonasfluorescens complex diversification of P. koreensis and P. mandelii from a commonancestor. We also hypothesize that the pathway was maintained in these Pseudomonasspp. by vertical transmission. In contrast, clade B contains clusters present in P. putidaand several species of the P. fluorescens complex in which there is no conservation ingenome localization, and the clusters are frequently associated with diverse elementsthat mediate horizontal gene transfer, such as transposases, integrases, and tRNAs (Fig.5B). This suggests different strategies to maintain koreenceine-type gene clusters indiverse Pseudomonas species.

Another Pseudomonas isolate (SWI36) was reported to inhibit B. cereus. Its activitywas dependent on a koreenceine-type gene cluster from clade B (18), but we foundthat the strain did not inhibit F. johnsoniae. Under certain conditions, P. koreensis

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 7

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

inhibited B. cereus, and the activity was dependent on the koreenceine cluster, sug-gesting a similarity with the SWI36 cluster (Fig. S9). Indeed, targeted metabolomicanalysis of Pseudomonas sp. SWI36 cell-free culture detected koreenceine C (Fig. S10).These data suggest that different koreenceine-like gene clusters in Pseudomonasgenomes have the capacity to synthesize koreenceine metabolites.

DISCUSSION

In this work, we aimed to understand the molecular basis for the growth inhibitionof F. johnsoniae by P. koreensis on a route to elucidating interactions within therhizosphere microbiome. We have shown that P. koreensis inhibits F. johnsoniae growththrough the production and secretion of novel secondary metabolites, koreenceine Ato D (metabolites 1 to 4), which have structural similarity to the plant metabolite�-coniceine. Based on the biosynthetic gene cluster identified through our geneticscreen, we propose a type II polyketide biosynthetic pathway for these bacterialalkaloids. Classical type II polyketide synthases are iterative heterodimer systems. It iscurrently unclear if the two heterodimer systems present in the biosynthetic cluster actin a modular or iterative manner, but the number of putative �-ketoacyl synthase genesin the pathway is consistent with modular biosynthesis. Plants also use a polyketidepathway mediated by an iterative type III polyketide synthase, providing a newexample of convergent evolution between these organisms for the synthesis of related

FIG 5 Phylogenetic analysis of the koreenceine biosynthetic locus and its distribution across bacteria. (A) ML phylogenetic tree estimated from the amino acidsequence of KecF and the corresponding structure of the koreenceine-like gene cluster present in each clade. (B) Schematic representation of thekoreenceine-like biosynthetic locus and its genomic context from several Pseudomonas spp. from clade A and clade B. Genes conserved in all genomes fromP. mandelii and P. koreensis are in dark gray, and variable or unique genes are in gray. Genes that likely experienced horizontal gene transfer events are in yellow.(C) Comparison of the phylogenies of kecF genes and their associated Pseudomonas genomes belonging to clade A kecF homologues. Both phylogeniescorrespond to ML analyses of the nucleotide sequence of the genes or the core genome. The dotted lines connect the cluster with its correspondingPseudomonas genome.

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 8

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

alkaloids. The tetrahydropyridine core of the koreenceine metabolites is found inwell-known plant alkaloids, such as the active cytotoxin �-coniceine from poisonhemlock (C. maculatum). Thus, koreenceine alkaloids may play roles in interbacterialand interdomain communication or inhibition that changes the rhizosphere commu-nity structure.

Members of the genus Pseudomonas are ubiquitous in nature and thrive in soil, onplants, and on moist surfaces. P. koreensis and other members of the Pseudomonasfluorescens complex are often studied for their capacity to colonize the rhizosphere andprotect plants from pathogens. Previous research demonstrated that P. fluorescenssuppresses plant disease through production of phenazine-1-carboxylic acid (PCA),which targets the fungal pathogen Gaeumannomyces graminis (20). P. fluorescens alsoproduces a suite of antimicrobial compounds, including 2,4-diacetylphloroglucinol,pyoluteorin, pyrrolnitrin, lipopeptides, and hydrogen cyanide (4), and members of theP. fluorescens complex also produce plant hormones, such as indole acetic acid andgibberellic acid, that stimulate plant growth. In this paper, we expand the knownrepertoire of metabolites from the P. fluorescens complex with discovery of koreenceineA to D. Unlike most of the P. fluorescens metabolites that inhibit fungal pathogens, thekoreenceines mediate interactions between P. koreensis and diverse members of theBacteroidetes, including F. johnsoniae, in a family-specific manner (6). Competitionbetween members of the P. fluorescens complex and Flavobacterium spp. in naturalsettings has been reported; in vivo studies showed a selective reduction of Flavobac-terium spp. in the Arabidopsis thaliana rhizosphere when Pseudomonas sp. CH267 wasadded to soil (21). Together, these results highlight the relevance of characterizingbacterium-bacterium interactions in the rhizosphere.

We identified a gene cluster necessary for the production of the koreenceinemetabolites. Other koreenceine-like clusters have been predicted to mediate (22) andothers are associated with (17, 18) antagonistic activity against diverse microorganisms.Thirty-five percent of P. koreensis and P. mandelii genomes in the NCBI database containthis gene cluster, and there are at least 160 Pseudomonas genomes harboring a relatedcluster, indicating the widespread nature of this pathway among Pseudomonas spp.Despite its ubiquity, structural characterization of the products of the biosynthetic genecluster remained undefined until this study.

Bioinformatic analysis of the proposed activities of the genes in the cluster enabledus to propose a biosynthetic pathway for the formation of a C14 polyketide with atetrahydropyridine-type ring from a noncanonical type II PKS system (Fig. 2). In bacteria,tetrahydropyridine-type rings could be derived from lysine cyclization (23), as observedin plants, or by a two-step reduction-transamination route of polyketide intermediates(24). We propose that the multidomain protein KecF directs both steps, reduction of theacyl intermediate to generate the acyl-aldehyde and transamination. This differs fromthe established route, in which the two activities are encoded by different genes, andthe reductase domain represents the terminal domain of a type I polyketide synthase(Fig. 4) (24). We propose that �-ketoacyl synthase(s) incorporates trans-2-decenoyl,decanoyl, or 3-hydroxy-decanoyl units to generate koreenceine A, B, or C, respectively.It is unclear if these acyl units are recruited as coenzyme A (CoA) esters from the�-oxidation pathway or as ACP esters from fatty acid synthesis, although the Rconfiguration in koreenceine D suggests substrate sampling from the fatty acid syn-thesis pool (i.e., fatty acyl-ACP).

Koreenceine A to C share structural features with �-coniceine, the metaboliteresponsible for the toxicity of the poison hemlock, a plant once used in death sentencesand the means by which Socrates took his own life after receiving such a sentence (399BCE). The structural similarity is the result of convergent evolution and suggestsadditional functionality of the metabolites. The koreenceines might play a protectiverole similar to that of the hemlock toxins, protecting plant roots from invertebrate pestsby poisoning their nervous systems (9), or the koreenceines might alter plant devel-opment, since �-coniceine is thought to be a plant hormone (9, 25). Thus, future work

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 9

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

will focus on characterizing the effect of koreenceine A to C on plant development andprotection.

MATERIALS AND METHODSBacterial strains and culture conditions. F. johnsoniae CI04, P. koreensis CI12, B. cereus UW85,

Pseudomonas sp. SWI36, Flavobacterium johnsoniae CI64, Chryseobacterium sp. CI02, Chryseobacterium sp.CI26, Sphingobacterium sp. CI01, and Sphingobacterium sp. CI48 were propagated on 1/10-strengthtryptic soy agar and grown in liquid culture in half-strength tryptic soy broth (TSB) at 28°C with vigorousshaking.

Production of root exudates and defined media. Soybean seeds were surface sterilized with 6%sodium hypochlorite for 10 min, washed with sterile deionized water, transferred to water agar plates,and allowed to germinate for 3 days in the dark at 25°C. Twelve seedlings were grown in a hydroponicsystem using 250 ml of modified Hoagland’s plant growth solution (26), which was collected after 10 daysof plant growth in an environmental chamber (12-h photoperiod, 25°C), filter sterilized, and stored at�20°C until used as root exudate. A defined medium was based on basal salt medium [1.77 g liter�1

Na2HPO4, 1.70 g liter�1 KH2PO4, 1.00 g liter�1 (NH4)2SO4, 0.16 g liter�1 MgCl2·6H2O, 5.00 mg liter�1

FeSO4·7H2O]. A carbon source (pyruvate, mannitol, or glucose) was added to a final concentration of4 mM. An amino acid mix of equal parts alanine, aspartate, leucine, serine, threonine, and valine wasadded to the root exudate or the defined medium at a final concentration of 6 mM. Individual aminoacids and N-methyl-DL-aspartate were also added to a final concentration of 6 mM.

P. koreensis mutant library generation by transposon mutagenesis. P. koreensis CI12 andEscherichia coli S17-1�pir with pSAM_BT20 (27) supplemented with ampicillin (100 �g ml�1) were firstgrown individually for 16 h in LB at 28°C and 37°C, respectively, with agitation. Cells were washed andresuspended in LB to an optical density at 600 nm of 2.0. One volume of E. coli S17-1�pir withpSAM_BT20 was mixed with two volumes of P. koreensis CI12. Cells were harvested (6,000 � g, 6 min),resuspended in 100 �l of LB, and spotted on LBA. Plates were incubated at 28°C for 16 h. Eachconjugation mixture was scraped off the plate and resuspended in 2.5 ml of LB, and 350-�l aliquots wereplated on LB containing gentamicin (50 �g ml�1) and chloramphenicol (10 �g ml�1) to select for P.koreensis CI12 transconjugants. Plates were incubated for 2 days at 28°C.

Genetic screen of P. koreensis mutants defective in inhibitory activity. P. koreensis CI12 mutantswere grown for 16 h in 96-deepwell plates filled with half-strength TSB, covered with sterile breathablesealing films, and incubated at 28°C with agitation. For each plate, the first well was inoculated withwild-type P. koreensis CI12 and the last well was left without P. koreensis CI12. F. johnsoniae CI04 wasgrown and washed as described above. Root exudate was inoculated with �107 F. johnsoniae CI04 cellsper ml, and 200-�l aliquots were added to each well of 96-well microplates. Two �l from each mutantP. koreensis CI12 culture was transferred to the corresponding wells on the microplates, which were thencovered by sealing films and incubated at 28°C with slight agitation for 2 days. Five �l from each wellwas then spotted on Casitone-yeast extract agar (CYE) (10 g liter�1 Casitone, 5 g liter�1 yeast extract,8 mM MgSO4,10 mM Tris buffer, 15 g liter�1 agar) containing kanamycin (10 �g ml�1) to select for F.johnsoniae, and plates were incubated at 28°C for 2 days. In the wild-type control, no F. johnsoniaecolonies were detected in the spot. Mutants that did not inhibit F. johnsoniae, which grew in the spotfrom mutant cultures, were streaked on a second plate for further analysis. The loss of inhibitory activityof candidate P. koreensis mutants was verified in a second coculture, and mutant growth was thencompared to wild-type growth to rule out candidates that failed to inhibit F. johnsoniae due to their owngrowth deficiency.

Location transposons in P. koreensis mutants defective in F. johnsoniae inhibition. For eachmutant, 1 ml of liquid culture grown for 16 h was harvested (6,000 � g, 6 min), and cells were resus-pended in 400 �l of TE (10 �M Tris-HCl, pH 7.4; 1 �M EDTA, pH 8.0). Samples were boiled for 6 min andcentrifuged (6,000 � g, 6 min), and 2 �l of supernatant was used as a template for DNA amplification.Transposon locations were determined by arbitrarily primed PCR, which consisted of a nested PCR usingfirst-round primer GenPATseq1 and either AR1A or AR1B and second-round primer GenPATseq2 and AR2(Table 2). PCR products from the second round were purified by gel extraction (QIAquick gel extractionkit; Qiagen) and then sequenced using primer GenPATseq2. Genomic regions flanking the transposonswere identified by comparing the sequences to the genome using the nucleotide BLAST alignment tool.

Chromosomal deletion of the koreenceine gene cluster in P. koreensis. The koreenceine clusterwas deleted by allelic exchange and replaced with a tetracycline resistance cassette. The kecA-kecKdeletion cassette was constructed by a modified version of an overlap extension (OE) PCR strategy.Fragments 1 kb upstream and 1 kb downstream of the kecA to kecK genes were amplified using primersmutSGCA_For/mutSGCA_Rev and mutSGCB_For/mutSGCB_Rev respectively (Table 2). The PCR productswere cloned in pENTR/d-TOPO, generating pkecA-K_ENTR. Primers mutSGCA_Rev and mutSGCB_Forwere designed to include a KpnI site in their overlapping region to allow introduction of a resistancegene. A tetracycline resistance cassette was amplified from pACYC184 using primers TetA_For/TetA_Rev,which contain KpnI sites in the 5= region, and cloned into pENTR/d-TOPO to generate pTetA_ENTR. Amob element was amplified from pJN105 using primers pJN105Mob_For/pJN105Mob_Rev (Table 2), inwhich an AscI site in the 5= region was added, and cloned in pENTR/d-TOPO, generating pmo-b_ENTR. The tetracycline resistance cassette was recovered from pTetA_ENTR using KpnI and clonedbetween the region upstream and downstream of pkecA-K_ENTR, and the mob element wasrecovered from pmob_ENTR using AscI and cloned into an AscI site in the pENTR backbone, generatingpkecA-K_TetA_mob_ENTR. Conjugation mixtures of P. koreensis CI12 and E. coli S17-1�pir carrying thepkecA-K_TetA_mob_ENTR vector were prepared by following the procedure for transposon mutant

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 10

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

generation. Double recombinant P. koreensis CI12 transconjugants were selected by their ability to growon tetracycline (10 �g ml�1) and inability to grow on kanamycin (50 �g ml�1). The kekA-kekK deletionmutant was confirmed by PCR using primers mutSGCA_For and mutSGCB_Rev. The kekA-kekK deletionmutant was further confirmed by evaluating growth of F. johnsoniae, and other members of theBacteroidetes, in its presence.

General information for the analysis and identification of metabolites. 1H and 2D (gCOSY,gHSQC, and gHMBC) NMR spectra were obtained on an Agilent (USA) 600-MHz NMR spectrometerwith a cold probe, and the chemical shifts were recorded as � values (ppm) with methanol-d4 as thestandard NMR solvent. Materials were routinely analyzed on an Agilent 6120 single-quadrupole liquidchromatography-mass spectrometry (LC-MS) system (Phenomenex Kinetex C18 column, 250 by 4.6 mm,5 �m; flow rate, 0.7 ml min�1; mobile phase composition, H2O and acetonitrile (ACN) containing 0.1%trifluoroacetic acid (TFA); method, 0 to 30 min with 10% to 100% ACN, hold for 5 min with 100% ACN,hold for 1 min with 100% to 10% ACN). High-resolution electrospray ionization mass spectrometry(HR-ESI-MS) data were obtained using an Agilent iFunnel 6550 QTOF mass spectrometer fitted with anESI source coupled to an Agilent (USA) 1290 Infinity HPLC system. Open-column chromatography wascarried out on a Waters Sep-Pak Vac 35-cc (10-g) C18 column. Metabolite isolations were performed usingan Agilent (USA) Prepstar HPLC system with an Agilent (USA) Polaris C18-A 5-�m (21.2- by 250-mm)column, a Phenomenex (USA) Luna C18(2) (100-Å) 10-�m (10.0- by 250-mm) column, a Phenomenex(USA) Luna C18(2) (100-Å) 10-�m (10.0- by 250-mm) column, and an Agilent Polaris 5 Amide-C18

(250- by 10.0-mm) column.Isolation of metabolites. P. koreensis CI12 was grown in defined medium with pyruvate as the

carbon source and supplemented with the amino acid mix or 3 mM glutamate for 3 days. Crude extractwas generated by liquid-liquid extraction using one volume of 2-butanol per one volume of filtersupernatant and dried by rotary evaporation. The crude extracts (400 mg) from the 12-liter culturesupernatant were resuspended in water and methanol (1:1 ratio), adsorbed onto Celite 110, and dried byrotary evaporation. The resulting powdery materials were loaded on the Waters Sep-Pak Vac 35-cc (10-g)C18 cartridge, and the metabolites were separated by solvent fractionation, eluting with a step gradientfrom 20% to 100% aqueous methanol to yield five subfractions (20%, 40%, 60%, 80%, and 100%methanol containing 0.1% TFA). Reverse-phase LC-MS analysis (10% to 100% aqueous acetonitrile in0.1% TFA, 30-min gradient) revealed that the 60% fraction included both molecules 1 and 2, and thefraction was dried under reduced pressure. This fraction (60 mg) was then separated by reverse-phaseHPLC equipped with an Agilent Polaris C18-A 5-�m (21.2- by 250-mm) column with an isocratic solventsystem (50% acetonitrile in water and 0.1% TFA over 20 min, 8 ml min�1, 1-min fraction collectioninterval). Compound 1 from the pooled fraction (11 and 12) (retention time [tR] � 25.3 min, 0.2 mg) waspartially purified over the Phenomenex Luna C18(2) 10-�m (10.0- by 250-mm) column with a lineargradient elution (20% to 80% acetonitrile in water and 0.1% TFA over 30 min). The combined HPLCfraction (11 and 12) was subsequently purified by reverse-phase HPLC [Phenomenex Luna C18(2) 10-�m(10- by 250-mm) column] with a linear gradient elution (20% to 80% acetonitrile in water and 0.1% TFAover 30 min) to yield pure compound 2 (tR � 25.8 min, 1.2 mg). Compound 4 was detected in the 80%aqueous methanol Sep-Pak fraction and was separated over an Agilent Polaris C18-A 5-�m (21.2- by 250-mm) column (flow rate, 8.0 ml/min; gradient elution, 10% to 100% aqueous acetonitrile in 0.1% TFA for30 min, 1-min fraction collection). HPLC fraction 24 was then separated over the Phenomenex Luna C18(2)10-�m (10- � 250-mm) column with 50% to 100% acetonitrile in water containing 0.1% TFA over 30 minat a flow rate of 4 ml min�1, followed by being subjected to an Agilent Polaris 5 Amide-C18 (250- by 10-mm) column with the same elution system (flow rate, 4 ml min�1; purification method, 50% to 100%acetonitrile in water containing 0.1% TFA over 30 min) to yield pure compound 4 (tR � 9.43 min, 0.7 mg).

(i) (E)-6-(non-1-en-1-yl)-2,3,4,5-tetrahydropyridine (compound 1). Colorless solid; 1H NMR(CD3OD, 600 MHz) � 7.21 to 7.12 (1H, m, H-8), 6.35 (1H, d, J � 16.0 Hz, H-7), 3.59 (2H, m, H-2), 2.91 (2H,m, H-5), 2.30 (2H, m, H-9), 1.77 (2H, m, H-3), 1.71 (2H, m, H-4), 1.42 (2H, m, H-10), 1.27 to 1.20 (8H, m, H-11,H-12, H-13, H-14), 0.84 (3H, t, J � 7.0 Hz, H-15); HR-ESI-QTOF-MS [M � H]� m/z 208.2067 (calculated forC14H26N, 208.2065).

(ii) 6-Nonyl-2,3,4,5-tetrahydropyridine (compound 2). Colorless solid; 1H NMR (CD3OD, 600 MHz)� 3.53 (2H, t, J � 5.5 Hz, H-2), 2.72 (2H, t, J � 6.1 Hz, H-5), 2.53 (2H, m, H-7), 1.73 (2H, m, H-3), 1.68 (2H, m,

TABLE 2 Primers used in this study

Name Sequence

GenPATseq1 CTTGGATGCCCGAGGCATAGGenPATseq2 CTGTACAAAAAAACAGTCATAACAAGCCATGAR1A GGCCACGCGTCGACTAGTACNNNNNNNNNNGTAATAR1B GGCCACGCGTCGACTAGTACNNNNNNNNNNGATGCAR2 GGCCACGCGTCGACTAGTACmutSGCA_For CACCCGCAAGCCTGCAATAGACGGACmutSGCA_Rev CCTGTCGTCTCAGGAAAGGTGCGGTACCTTCTATCTCCCTATATGTCGTGACmutSGCB_For GTCACGACATATAGGGAGATAGAAGGTACCGCACCTTTCCTGAGACGACAGGmutSGCB_Rev GCACCTGACATTCGTCTATCCGATCTetA_For CACCGGTACCTCCTCCAAGCCAGTTACCTCGGTetA_Rev GGTACCTGCTCAGGTCGCAGACGTTTTGpJN105Mob_For TAGGCGCGCCTGTGGTCAAGCTCGTGGGCpJN105Mob_Rev CACCGGCGCGCCCAATTCGTTCAAGCCGAGATCGGC

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 11

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

H-4), 1.54 (2H, dt, J � 14.6, 6.8 Hz, H-8), 1.28-1.15 (12H, m, H-9, H-10, H-11, H-12, H-13, H-14), 0.83 (3H, t,J � 7.0 Hz, H-15), 13C NMR (CD3OD, 125 MHz) � 192.1 (C-6), 44.4 (C-2), 37.7 (C-7), 31.6 (C-13), 29.4 (C-5),28.0 to 29.0 (C-9, C-10, C-11, C-12), 25.5 (C-8), 22.5 (C-14), 19.2 (C-3), 16.8 (C-4), 14.4 (C-15); HR-ESI-QTOF-MS [M � H]� m/z 210.2224 (calculated for C14H28N, 210.2222).

(iii) (R)-N-(4-Chlorobutyl)-3-hydroxydecanamide (compound 4). Colorless solid; 1H NMR (CD3OD,600 MHz) � 3.94 (1H, m, H-3), 3.59 (2H, t, J � 6.5 Hz, H-4=), 3.20 (2H, m, H-1=), 2.27 (2H, m, H-2), 1.78 (2H,m, H-3=), 1.63 (2H, dt, J � 14.3, 7.0 Hz, H-2=), 1.43 (2H, m, H-4), 1.36 to 1.25 (10H, m, H-5, H-6, H-7, H-8, H-9),0.90 (3H, t, J � 6.9 Hz, H-10), 13C NMR (CD3OD, 125 MHz) � 172.6 (C-1), 68.2 (C-3), 44.1 (C-4=), 43.8 (C-2),37.9 (C-1=), 36.8 (C-4), 31.8 (C-8), 29.5 (C-3=), 29.4 (C-5 or C-6, C-7), 26.3 (C-2=), 22.6 (C-5 or C-6, C-9), 13.4(C-10); HR-ESI-QTOF-MS [M � H]� m/z 278.1885 (calculated for C14H29ClNO2, 278.1887).

Determination of absolute configuration of metabolite 4. The absolute configuration of metabolite 4was determined using the modified Mosher’s method with R- and S-�-methoxy-(trifluoromethyl)phenylacetylchloride (MTPA-Cl) (28). Compound 4 (0.5 mg) was prepared in two vials (0.25 mg), and each sample wasdissolved in 250 �l of dried pyridine-d5 in vials purged with N2 gas. Dimethylaminopyridine (DMAP)(0.5 mg) was added to both vials, followed by the addition of 5 �l of S- and R-MTPA-Cl solution (2%,vol/vol) at room temperature. After 18 h, the reaction mixtures were dried under reduced pressure. 1HNMR spectra of the Mosher esters (S-MTPA ester and R-MTPA ester) were collected in methanol-d4, andthe chemical shift differences of the Mosher esters of compound 4 were calculated as Δ�S-R.

Characterization of Pseudomonas sp. SWI36. Pseudomonas sp. SWI36 and P. koreensis inhibitoryinteractions against B. cereus and F. johnsoniae were evaluated with a modified spread-patch method.Strains were grown separately for 20 h. One-ml aliquots of cultures of each strain were centrifuged(6,000 � g, 6 min) and resuspended in 1 ml of the same medium (undiluted cultures), and a 1:100 dilutionof B. cereus and F. johnsoniae was prepared in the same medium (diluted culture). Nutrient agar plateswere spread with 100 �l of either B. cereus or F. johnsoniae diluted cultures and spotted with 10 �l of theundiluted cultures of Pseudomonas sp. SWI36 and P. koreensis. Plates were then incubated at 28°C andinspected for zones of inhibition after 2 days. Crude extract of Pseudomonas sp. SWI36 and Pseudomonassp. SWI36 kecF::Tn culture in nutrient broth were prepared as described above. Extracted materials wereanalyzed on an LC-MS system consisting of a Thermo Fisher Scientific (Waltham, MA) Q Exactive Orbitrapmass spectrometer with an ESI source coupled to a Vanquish UHPLC (Thermo Accucore Vanquish C18

column, 100 by 2.1 mm, 1.5 �m; flow rate, 0.2 ml min�1; mobile phase composition, H2O and ACNcontaining 0.1% TFA; method, 0 to 1 min with 10% ACN, 1 to 4 min with 10% to 35% ACN, 4 to 12 minwith 35% to 70% ACN, 12 to 16 min with 70% to 98% ACN, 16- to 20-min hold with 98% ACN, 20 to21 min with 98% to 10% ACN, and 21 to 23 min with 10% ACN). MS1 scans were acquired with positiveionization over an m/z range of 188 to 1,275 with settings of 1e�6 AGC, 100-ms maximum integrationtime, and 70-K resolution.

Phylogenetic analysis. Genetic regions with homology to the koreenceine biosynthetic cluster wereidentified by BLAST alignment tools (29) using the P. koreensis CI12 KecF protein sequence in the NCBIdatabase. All of the KecF homologues identified were part of koreenceine-like clusters (genomesharboring these are listed in Table S1 in the supplemental material). Protein and nucleotide sequencealignments of kecF were performed with MAFFT, version 7 (30), and were manually adjusted using as aguide the residue-wise confidence scores generated by GUIDANCE2 (31). Best-fit models of amino acidor nucleotide replacement were selected. Evolutionary analyses were inferred by maximum likelihood(ML) methods conducted in MEGA X (32). The P. koreensis and P. mandelii phylogenomic reconstructionwas done by phylogenetic and molecular evolutionary (PhaME) analysis software (33). PhaME identifiedsingle-nucleotide polymorphisms from the core genome alignments, and the phylogenetic relationshipswere inferred by ML using FastTree. Phylogenetic trees were visualized using the interactive tree of life(iTOL) (34).

SUPPLEMENTAL MATERIALSupplemental material for this article may be found at https://doi.org/10.1128/AEM

.03058-18.SUPPLEMENTAL FILE 1, PDF file, 2.2 MB.

ACKNOWLEGMENTSWe are grateful to Sailendharan Sudakaran for discussing phylogenetic analysis. We

thank Hans Wildschutte for Pseudomonas sp. SWI36 and its mutant. This work wassupported by the Office of the Provost at Yale University, funding from the WisconsinAlumni Research Foundation through the University of Wisconsin–Madison Office ofthe Vice Chancellor for Research and Graduate Education, and NSF grant MCB-1243671.

REFERENCES1. Berg G, Grube M, Schloter M, Smalla K. 2014. Unraveling the plant

microbiome: looking back and future perspectives. Front Microbiol5:148. https://doi.org/10.3389/fmicb.2014.00148.

2. Bais HP, Weir TL, Perry LG, Gilroy S, Vivanco JM. 2006. The role of rootexudates in rhizosphere interactions with plants and other organisms.

Annu Rev Plant Biol 57:233–266. https://doi.org/10.1146/annurev.arplant.57.032905.105159.

3. Arshad M, Frankenberger WT, Jr. 1998. Plant growth-regulating sub-stances in the rhizosphere: microbial production and functions. AdvAgron 62:45–151.

Lozano et al. Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 12

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

4. Handelsman J, Stabb EV. 1996. Biocontrol of soilborne plant pathogens.Plant Cell 8:1855–1869. https://doi.org/10.1105/tpc.8.10.1855.

5. Bulgarelli D, Schlaeppi K, Spaepen S, Ver Loren van Themaat E, Schulze-Lefert P. 2013. Structure and functions of the bacterial microbiota ofplants. Annu Rev Plant Biol 64:807– 838. https://doi.org/10.1146/annurev-arplant-050312-120106.

6. Lozano GL, Bravo JI, Garavito Diago MF, Park HB, Hurley A, Peterson SB,Stabb EV, Crawford JM, Broderick NA, Handelsman J. 2019. IntroducingTHOR, a model microbiome for genetic dissection of community behav-ior. mBio 10:e02846-18. https://doi.org/10.1128/mBio.02846-18.

7. Valentini M, García-Mauriño SM, Pérez-Martínez I, Santero E, Canosa I,Lapouge K. 2014. Hierarchical management of carbon sources is regu-lated similarly by the CbrA/B systems in Pseudomonas aeruginosa andPseudomonas putida. Microbiology 160:2243–2252. https://doi.org/10.1099/mic.0.078873-0.

8. Nishijyo T, Haas D, Itoh Y. 2001. The CbrA-CbrB two-component regu-latory system controls the utilization of multiple carbon and nitrogensources in Pseudomonas aeruginosa. Mol Microbiol 40:917–931. https://doi.org/10.1046/j.1365-2958.2001.02435.x.

9. Vetter J. 2004. Poison hemlock (Conium maculatum L.). Food ChemToxicol 42:1373–1382. https://doi.org/10.1016/j.fct.2004.04.009.

10. Leete E. 1964. Biosynthesis of the hemlock alkaloids. The incorporationof acetate-1-C14 into coniine and conhydrine. J Am Chem Soc 86:2509 –2513. https://doi.org/10.1021/ja01066a039.

11. Hotti H, Seppänen-Laakso T, Arvas M, Teeri TH, Rischer H. 2015.Polyketide synthases from poison hemlock (Conium maculatum L.). FEBSJ 282:4141– 4156. https://doi.org/10.1111/febs.13410.

12. Roberts MF. 1971. The formation of �-coniceine from 5-ketooctanal by atransaminase of Conium maculatum. Phytochemistry 10:3057–3060.https://doi.org/10.1016/S0031-9422(00)97351-5.

13. Roberts MF. 1981. Enzymic synthesis of �-coniceine in Conium macula-tum chloroplasts and mitochondria. Plant Cell Rep 1:10 –13. https://doi.org/10.1007/BF00267647.

14. Schirmer A, Rude MA, Li X, Popova E, del Cardayre SB. 2010. Microbialbiosynthesis of alkanes. Science 329:559 –562. https://doi.org/10.1126/science.1187936.

15. Blitzke T, Porzel A, Masaoud M, Schmidt J. 2000. A chlorinated amide andpiperidine alkaloids from Aloe sabaea. Phytochemistry 55:979 –982.https://doi.org/10.1016/S0031-9422(00)00269-7.

16. Abe I, Morita H. 2010. Structure and function of the chalcone synthasesuperfamily of plant type III polyketide synthases. Nat Prod Rep 27:809 – 838. https://doi.org/10.1039/b909988n.

17. Chung EJ, Lim HK, Kim J-C, Choi GJ, Park EJ, Lee MH, Chung YR, Lee S-W.2008. Forest soil metagenome gene cluster involved in antifungal activ-ity expression in Escherichia coli. Appl Environ Microbiol 74:723–730.https://doi.org/10.1128/AEM.01911-07.

18. Davis E, Sloan T, Aurelius K, Barbour A, Bodey E, Clark B, Dennis C, DrownR, Fleming M, Humbert A, Glasgo E, Kerns T, Lingro K, McMillin M, MeyerA, Pope B, Stalevicz A, Steffen B, Steindl A, Williams C, Wimberley C,Zenas R, Butela K, Wildschutte H. 2017. Antibiotic discovery throughoutthe Small World Initiative: a molecular strategy to identify biosyntheticgene clusters involved in antagonistic activity. MicrobiologyOpen6:e00435. https://doi.org/10.1002/mbo3.435.

19. Garrido-Sanz D, Meier-Kolthoff JP, Göker M, Martín M, Rivilla R, Redondo-

Nieto M. 2016. Genomic and genetic diversity within the Pseudomonasfluorescens complex. PLoS One 11:e0150183. https://doi.org/10.1371/journal.pone.0150183.

20. Thomashow LS, Weller DM. 1988. Role of a phenazine antibiotic fromPseudomonas fluorescens in biological control of Gaeumannomycesgraminis var. tritici. J Bacteriol 170:3499 –3508. https://doi.org/10.1128/jb.170.8.3499-3508.1988.

21. Haney CH, Samuel BS, Bush J, Ausubel FM. 2015. Associations withrhizosphere bacteria can confer an adaptive advantage to plants.NPLANTS 1:15051. https://doi.org/10.1038/nplants.2015.51.

22. Martins dos Santos VAP, Timmis KN, Tümmler B, Weinel C. 2004.Genomic features of Pseudomonas putida strain KT2440, p 77–112. InRamos JL (ed), Pseudomonas. Springer US, Boston, MA.

23. Bis DM, Ban YH, James ED, Alqahtani N, Viswanathan R, Lane AL. 2015.Characterization of the nocardiopsin biosynthetic gene cluster revealssimilarities to and differences from the rapamycin and FK-506 pathways.Chembiochem 16:990 –997. https://doi.org/10.1002/cbic.201500007.

24. Peng H, Wei E, Wang J, Zhang Y, Cheng L, Ma H, Deng Z, Qu X. 2016.Deciphering piperidine formation in polyketide-derived indolizidinesreveals a thioester reduction, transamination, and unusual imine reduc-tion process. ACS Chem Biol 11:3278 –3283. https://doi.org/10.1021/acschembio.6b00875.

25. Reynolds T. 2005. Hemlock alkaloids from Socrates to poison aloes.Phytochemistry 66:1399–1406. https://doi.org/10.1016/j.phytochem.2005.04.039.

26. Hoagland DR, Arnon DI. 1950. The water-culture method for growingplants without soil. Calif Agric Exp Stat Circ 347:1–32.

27. Sivakumar R, Ranjani J, Vishnu US, Jayashree S, Lozano GL, Miles J,Broderick N, Guan C, Gunasekaran P, Handelsman J, Rajendhran J. 2019.Evaluation of InSeq to identify genes essential for Pseudomonas aerugi-nosa PGPR2 corn root colonization. G3 (Bethesda) 9:651– 661. https://doi.org/10.1534/g3.118.200928.

28. Hoye TR, Jeffrey CS, Shao F. 2007. Mosher ester analysis for the deter-mination of absolute configuration of stereogenic (chiral) carbinol car-bons. Nat Protoc 2:2451–2458. https://doi.org/10.1038/nprot.2007.354.

29. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic localalignment search tool. J Mol Biol 215:403– 410. https://doi.org/10.1016/S0022-2836(05)80360-2.

30. Katoh K, Standley DM. 2013. MAFFT multiple sequence alignment soft-ware version 7: improvements in performance and usability. Mol BiolEvol 30:772–780. https://doi.org/10.1093/molbev/mst010.

31. Sela I, Ashkenazy H, Katoh K, Pupko T. 2015. GUIDANCE2: accuratedetection of unreliable alignment regions accounting for the uncertaintyof multiple parameters. Nucleic Acids Res 43:W7–14. https://doi.org/10.1093/nar/gkv318.

32. Kumar S, Stecher G, Li M, Knyaz C, Tamura K. 2018. MEGA X: molecularevolutionary genetics analysis across computing platforms. Mol Biol Evol35:1547–1549. https://doi.org/10.1093/molbev/msy096.

33. Ahmed SA, Lo C-C, Li P-E, Davenport KW, Chain PSG. 2015. From rawreads to trees: whole genome SNP phylogenetics across the tree of life.bioRxiv https://doi.org/10.1101/032250.

34. Letunic I, Bork P. 2016. Interactive tree of life (iTOL) v3: an online tool forthe display and annotation of phylogenetic and other trees. NucleicAcids Res 44:W242–W245. https://doi.org/10.1093/nar/gkw290.

Bacterial Alkaloids Mediate Microbial Interactions Applied and Environmental Microbiology

May 2019 Volume 85 Issue 10 e03058-18 aem.asm.org 13

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from